Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ReChorus2.0: A Modular and Task-Flexible Recommendation Library

Jiayu Li, Hanyu Li, Zhiyu He, Weizhi Ma, Peijie Sun, Min Zhang, Shaoping Ma

TL;DR

ReChorus2.0 addresses the fragmentation in recommender-system research tooling by delivering a modular, task-flexible library that supports Top-k recommendation, impression-based ranking/reranking, and CTR prediction within a unified framework. Built as an extension of ReChorus1.0, it introduces enhanced input formats, context-aware and re-ranking capabilities, and customizable candidate sets through a four-part design of Input, Reader, Model, and Runner, enabling a single model to support multiple tasks. The framework embodies up to 37 models across general, sequential, re-rank, and context-aware families and provides task-specific readers and runners to streamline training and evaluation. Empirical results on datasets like MIND and MovieLens-1M demonstrate favorable performance improvements from reranking and context-aware modeling, highlighting the practical impact for researchers seeking flexible experimentation and robust reproducibility. Overall, ReChorus2.0 offers researchers a powerful, extensible tool to assemble diverse recommendation experiments, aligning closely with real-world problems and aiming to accelerate methodological and data-driven insights in the field.

Abstract

With the applications of recommendation systems rapidly expanding, an increasing number of studies have focused on every aspect of recommender systems with different data inputs, models, and task settings. Therefore, a flexible library is needed to help researchers implement the experimental strategies they require. Existing open libraries for recommendation scenarios have enabled reproducing various recommendation methods and provided standard implementations. However, these libraries often impose certain restrictions on data and seldom support the same model to perform different tasks and input formats, limiting users from customized explorations. To fill the gap, we propose ReChorus2.0, a modular and task-flexible library for recommendation researchers. Based on ReChorus, we upgrade the supported input formats, models, and training&evaluation strategies to help realize more recommendation tasks with more data types. The main contributions of ReChorus2.0 include: (1) Realization of complex and practical tasks, including reranking and CTR prediction tasks; (2) Inclusion of various context-aware and rerank recommenders; (3) Extension of existing and new models to support different tasks with the same models; (4) Support of highly-customized input with impression logs, negative items, or click labels, as well as user, item, and situation contexts. To summarize, ReChorus2.0 serves as a comprehensive and flexible library better aligning with the practical problems in the recommendation scenario and catering to more diverse research needs. The implementation and detailed tutorials of ReChorus2.0 can be found at https://github.com/THUwangcy/ReChorus.

ReChorus2.0: A Modular and Task-Flexible Recommendation Library

TL;DR

ReChorus2.0 addresses the fragmentation in recommender-system research tooling by delivering a modular, task-flexible library that supports Top-k recommendation, impression-based ranking/reranking, and CTR prediction within a unified framework. Built as an extension of ReChorus1.0, it introduces enhanced input formats, context-aware and re-ranking capabilities, and customizable candidate sets through a four-part design of Input, Reader, Model, and Runner, enabling a single model to support multiple tasks. The framework embodies up to 37 models across general, sequential, re-rank, and context-aware families and provides task-specific readers and runners to streamline training and evaluation. Empirical results on datasets like MIND and MovieLens-1M demonstrate favorable performance improvements from reranking and context-aware modeling, highlighting the practical impact for researchers seeking flexible experimentation and robust reproducibility. Overall, ReChorus2.0 offers researchers a powerful, extensible tool to assemble diverse recommendation experiments, aligning closely with real-world problems and aiming to accelerate methodological and data-driven insights in the field.

Abstract

With the applications of recommendation systems rapidly expanding, an increasing number of studies have focused on every aspect of recommender systems with different data inputs, models, and task settings. Therefore, a flexible library is needed to help researchers implement the experimental strategies they require. Existing open libraries for recommendation scenarios have enabled reproducing various recommendation methods and provided standard implementations. However, these libraries often impose certain restrictions on data and seldom support the same model to perform different tasks and input formats, limiting users from customized explorations. To fill the gap, we propose ReChorus2.0, a modular and task-flexible library for recommendation researchers. Based on ReChorus, we upgrade the supported input formats, models, and training&evaluation strategies to help realize more recommendation tasks with more data types. The main contributions of ReChorus2.0 include: (1) Realization of complex and practical tasks, including reranking and CTR prediction tasks; (2) Inclusion of various context-aware and rerank recommenders; (3) Extension of existing and new models to support different tasks with the same models; (4) Support of highly-customized input with impression logs, negative items, or click labels, as well as user, item, and situation contexts. To summarize, ReChorus2.0 serves as a comprehensive and flexible library better aligning with the practical problems in the recommendation scenario and catering to more diverse research needs. The implementation and detailed tutorials of ReChorus2.0 can be found at https://github.com/THUwangcy/ReChorus.
Paper Structure (25 sections, 3 figures, 6 tables)

This paper contains 25 sections, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Reranking in Recommendations
  4. Context-aware Recommendation
  5. Libraries for Recommendation Algorithms
  6. The ReChorus2.0 Framework
  7. Supported Task Categories
  8. Overall Framework Design
  9. Data Readers
  10. Supported Models
  11. Training and Evaluation Runners
  12. Task-based Usage Guidelines
  13. Impression-based Ranking/Re-ranking
  14. CTR Prediction Task
  15. Context-aware Top-k Recommendation Task
  16. ...and 10 more sections

Figures (3)

  • Figure 1: The overall framework of ReChorus 2.0 (a) and some task implementation examples (b). ReChorus2.0 can fit various recommendation tasks by flexible assembly of four parts: input data, reader, model, and runner. Corresponding modules are automatically assembled after users specify the model and task.
  • Figure 2: Examples of data input formats supported by ReChorus2.0. Note that impression_id is only necessary when conducting impression-based tasks in Figure(b).
  • Figure 3: Examples of configurations to run different tasks with various types of recommenders.